The potential of uncertainty tolerance measures to identify specific needs of first-year medical residents: an exploratory study Authors: Pål Gulbrandsen, Jūratė Šaltytė Benth, Eirik Hugaas Ofstad, Paul K J Han Published: 2025 ABSTRACT AbstractBackgroundWe aimed to explore 1st year residents’ responses to five different uncertainty tolerance (UT) measures, varying in their specificity to physicians and medical practice, and to compare their ability to classify residents according to their level of UT in this population.Methods202 residents beginning training at two Norwegian hospitals in 2021–2023 completed the Physicians’ Reaction to Uncertainty (PRU), a physician-specific UT measure, as well as 4 non-specific measures: Intolerance of Uncertainty Scale (IUS-12), Tolerance for Ambiguity (TFA), Pearson’s Risk Attitude (PRA), and Ambiguity Aversion in Medicine (AAM). We described psychometric properties using exploratory factor analysis (EFA). We then applied hierarchical cluster analysis to assess the utility of the PRU and the 4 non-physician-specific measures in classifying residents.ResultsResidents’ mean length of clinical work experience was 9.6 months (SD 7.6). 137 (68.2%) were female, 125 (62.2%) went to medical schools in Scandinavian-speaking countries. The exploratory factor analysis of all measures together revealed five factors. Factor 1 consisted of all PRU items and four items from TFA and explained 50.4% of the variance. In step 1, the cluster analysis based on PRU alone produced three stable clusters, with high (n = 58, 29%), intermediate (n = 108, 54%), or low (n = 36, 18%) UT. In step 2, a cluster analysis based on all measures produced three clusters, with high (n = 81, 40%), intermediate (n = 67, 33%), or low (n = 54, 27%) UT. Of the 58 with high UT according to PRU, three (5%) had intermediate UT using all instruments combined. Of the 36 with low UT according to PRU, 12 (33%) had high or intermediate UT using all instruments combined. Female residents and residents who studied medicine outside Scandinavia displayed lower UT.ConclusionPRU appears to be a reliable, discriminatory, and feasible measure for identifying junior physicians who are predisposed to stress or other aversive psychological reactions to the uncertainties of clinical care. High UT measured by PRU indicates high UT according to other measures, while low UT according to PRU is less indicative. Hence, non-physician-specific UT measures appear to assess specific dimensions of UT that could assist supervisors and residents understand different responses to uncertainty.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12909-025-08449-0.Keywords: Uncertainty tolerance, Residents, Physicians, Norway Background Uncertainty tolerance (UT) is a trait-level characteristic that influences how individuals manage uncertainty—the subjective perception of ignorance [1], a typically aversive, undesirable state that leads to a variety of negative psychological responses [2]. Individual differences in UT predict the extent to which people exhibit negative vs. positive responses to uncertainty [2], and have been measured among both physicians and physician-trainees [3, 4]. Low UT has been shown to predict several negative responses to uncertainty, including lower levels of confidence with various aspects of clinical practice, burnout [5] and diminished emotional well-being, and suboptimal clinical decisions and behaviors including overuse of healthcare resources [3, 6]. UT thus has important implications for medical practice and education and may be valuable to assess, particularly because evidence suggests it may not be fixed and immutable, but alterable by clinical training and experience [7, 8]. Assessing trainees’ UT may thus serve valuable functions in medical education, including “screening” for individuals who are predisposed to negative psychological responses (e.g., diminished emotional well-being) to the uncertainties of clinical care, and who may thus benefit from specific coaching, supervision, or other interventions. Currently, however, the assessment of UT among physicians and physician-trainees is limited by substantial variability in its measurement. Numerous UT measures have been used in this population, but they differ in various ways. One measure, the Physicians’ Reactions to Uncertainty (PRU) scale, assesses physicians’ tolerance of various uncertainties in medical practice. Other measures are not physician-specific but are more specific in focusing on tolerance of particular uncertainties in life. Variability in both the physician-specificity and uncertainty-specificity of existing UT measures raises questions about the underlying constructs assessed by different measures, and their ability to distinguish between individual physicians and physician-trainees with differing levels of UT. We conducted an exploratory study aimed at generating preliminary answers to these questions. In this study, we administered five brief UT measures—the physician-specific PRU and four non-physician-specific but uncertainty-specific UT measures—to a group of physicians at commencement of their 1 st year of residency in Norway, and conducted analyses with two objectives: to compare the PRU and non-physician-specific UT measures in terms of the constructs they assess and their performance in classifying physicians according to their level of UT, in order to assess the potential value of adding non-physician specific measures to the PRU in medical education. Methods Study sample We collected questionnaires with UT measures and separate self-reported demographic information at introductory meetings before residency in two Norwegian hospitals. A code secured that only PG and EHO could identify physicians in case of withdrawal from the study. More than 95% of the audiences completed questionnaires. Residency starts in February or August. Questionnaires were distributed at five introductory meetings between February 2021 and February 2023; 202 questionnaires were completed. The study was exempt from review by the Regional Medical Ethics Committee of Southeast Norway, application #273,339. The Data Protection Officer at Akershus University Hospital has approved the way we store data (21/05892-6). UT measures Our selection of UT measures was guided by (a) the theoretical goal of assessing tolerance of uncertainty pertaining to different issues and arising from different identifiable sources, and (b) the practical goal of identifying measures that would be feasible to implement in medical education settings. We therefore selected five relatively brief existing measures that assessed tolerance of uncertainty about different issues (medical vs. non-medical uncertainties, experienced by physicians vs. laypersons) and arising from different sources (probability, ambiguity, complexity) [7]. The Physicians’ Reactions to Uncertainty (PRU) scale, the only physician-specific instrument in this study, is a validated 15-item measure that assesses tolerance of uncertainty pertaining to medical issues and experienced by physicians [9, 10]. For the current study, we used only the 5-item Anxiety Due to Uncertainty subscale of the measure [10], which specifically assesses physicians’ feelings of distress or discomfort with uncertainty in patient care. We excluded the “Concern about bad outcomes,” “Reluctance to disclose uncertainty to patients,” and “Reluctance to disclose mistakes to physicians” subscales, given that these items assess uncertainty responses that are likely influenced by contextual factors (e.g., institutional culture, local practice norms regarding malpractice liability and disclosure of medical errors). PRU is generic, neither explicitly nor implicitly focusing on any single type or source of uncertainty. Items are scored on a six-point Likert scale “Strongly disagree” (1) to “Strongly agree” (6), one item is reversed. The Intolerance of Uncertainty Scale (IUS-12) is a 12-item extensively validated one-dimensional measure widely used in clinical psychology research in both clinical and non-clinical populations [11, 12]. The IUS-12 is not specific to physicians or the medical domain and does not explicitly focus on any single source or issue of uncertainty. The scale consists of two dimensions, Prospective Anxiety and Inhibitory Anxiety; however for this analysis we utilized the entire scale. Each item is scored on five-point Likert scale from “Not at all characteristic of me” (1) to “Entirely characteristic of me” (5). Greater intolerance of uncertainty as assessed by the IUS-12 has been shown to predict various mental health outcomes including pathological levels of anxiety and depression as well as fear of negative evaluation and social distress [13]. The Tolerance for Ambiguity (TFA) scale is a 7-item measure to assess general tolerance of uncertainty in life [14, 15]. It is not specific to physicians or the medical domain and does not explicitly focus on tolerance of any single type of uncertainty; however, its items arguably focus primarily on uncertainty arising from complexity and ambiguity as opposed to probability [8, 16]. The items are scored on six-point Likert scale from “Strongly disagree” (1) to “Strongly agree” (6). TFA has been extensively validated and shown to consist of one dimension. Higher tolerance of ambiguity has been shown to predict medical students’ specialty choice and physician interest in ordering a genetic test of unknown clinical utility [14, 15]. The Pearson Risk Attitude (PRA) scale is a 6-item measure to assess tolerance of uncertainty arising specifically from risk, or probability [17]. It is not specific to physicians or the medical domain, and each item is scored on a six-point Likert scale “Strongly disagree” (1) to “Strongly agree” (6). However, as PRA is conceptually opposite to the other measures regarding risk, we inverted the items accordingly to avoid confusion in the tables. PRA is not specific to the medical domain. Lower risk tolerance among emergency medicine physicians has been shown to be associated with higher rates of admission for patients with acute chest pain [17]. The Ambiguity Aversion in Medicine (AAM) scale is a validated 6-item one-dimensional scale to assess tolerance of uncertainty arising specifically from ambiguity, conceptualized specifically in decision theory terms as uncertainty regarding the reliability, credibility, or adequacy of risk information [18]. It is specific to the medical domain, but not to physicians. It consists of five items, each scored on a six-point Likert scale “Strongly disagree” (1) to “Strongly agree” (6), two of them inversely scored. The AAM is specific to ambiguity in the medical domain; however originally designed for laypersons but used in past studies of physicians and physician-trainees [8, 18, 19], the scale was adapted for the current study by omitting a single item, “Conflicting expert opinions about a medical test or treatment would lower my trust in the experts.” Higher AAM scores have been shown to predict lower interest among laypersons in a hypothetical ambiguous cancer screening test, and pessimistic appraisals of the benefits and harms of cancer screening interventions [18, 20]. All instruments were translated to Norwegian by a Norwegian fully fluent in English. The Norwegian versions were back-translated by two native English speakers fluent in Norwegian or with good working knowledge of Norwegian. Translators discussed the Norwegian versions and reached agreement easily. Note that high scores on all measures indicate low UT. Statistical analysis We described the data by means and standard deviations (SDs) or frequencies and percentages, as appropriate. Although the measure items were recorded on ordinal (Likert) scales (1–6 or 1–5), we treated them as continuous variables for descriptive purposes and presented them as medians and min-max values, and as means and SDs. This enables easier comparison with similar studies that often treat the Likert scale as continuous, as supported by Norman [21]. Missing values were few (ranging 1–4 per item and constituting only 1.5% (n = 109) of data set). They were imputed with median values of the item for further analyses. To assess the dimensionality of the questionnaire, we applied exploratory factor analysis (EFA). The reason for applying EFA instead of confirmatory factor analysis, is that although each UT measure has a clear structure, cross-correlations among items might be present which makes it difficult to formulate a hypothesis about the number of dimensions. Due to the ordinal nature of the items, we calculated Spearman’s correlations among all pairs first. Next, we used the correlation matrix as input for the EFA. We applied three factor extraction methods, iterated principal factors, principal factors and principal-component factors, along with varimax and promax rotation methods, and screen-test in combination with Kaiser’s criterion of eigenvalue > 1 followed by parallel analysis. We assessed internal consistency of the identified dimensions by Cronbach’s alpha [22]. Since the Cronbach’s alpha’s strict assumptions are often violated, we also report the total omega coefficient [23] which relies on more relaxed assumptions and thus is often a more suitable measure of internal consistency. According to Nunnally & Bernstein, values above 0.7 indicate acceptable internal consistency [24]. Attempting to identify homogeneous groups of physicians using questionnaire responses we applied agglomerative hierarchical cluster analysis [25] with Ward Linkage method [26] based on squared Euclidean distances. We chose the hierarchical cluster analysis due to the exploratory nature of the study. This approach is not limited by a predefined number of clusters and often provides insights into multiple levels of clustering and underlying patterns. The dendrogram was visually inspected to help determine the number of clusters, taking also into account the size of each group and between-cluster similarities across different solutions. The average silhouette scores were presented. Also, Kendall’s W (coefficient of concordance) was calculated to explore the within-cluster consistency. We compared the single items and physician characteristics in the identified groups by ANOVA and χ2-test, as appropriate. We conducted the cluster analyses in two steps that differed in the scales analyzed: (1) PRU alone and (2) all scales combined. We reasoned that the PRU scale would likely perform best in classifying trainees given its physician-specific nature, and that evaluating its performance both with and without the other, non-physician-specific scales would yield insights about the added value of these other scales. All tests were two-sided and results with p-values below 0.05 were considered statistically significant. Given the exploratory nature of our study, we have chosen not to adjust for multiple testing. Instead, we report all test results and thus ensure transparency, allowing readers to make their own judgment when interpreting the findings. All statistical analyses were performed in SPSS v29. STATA v18 was used to generate a Sankey diagram, which illustrates how different combinations of the UT measures performed in separating physicians into groups. Results Two thirds of the residents were female, and more than 60% graduated from Scandinavian speaking medical schools (Table 1). Mean work experience was 9.6 (SD 7.6) months, with substantial variation (0 to 45 months). Table 1Resident gender and contextual characteristics, n=202CharacteristicStatisticMonths of work experience n193 min, max0, 45 mean (SD)9.6 (7.6)Gender, n (%) n202 Female137 (68.2)Geographical region of medical university, n (%) n201 Scandinavia125 (62.2) Outside Scandinavia76 (37.8) Resident gender and contextual characteristics, n=202 Descriptive statistics for questionnaire items are presented in Table 2. The distribution was slightly skewed for most items. The correlation structure including all scale items (Supplementary Table S1) showed no clear dimensions. Items within the PRU scale showed the strongest intercorrelations, suggesting one dimension. Most IUS-12 scale items and some TFA and PRA items exhibited correlations with PRU scale items, supporting the use of EFA. Table 2Descriptive statistics of items in included instruments, n=202ItemMedian (min-max)Mean (SD)I. Intolerance of Uncertainty Scale (IUS) (Carleton et al., 2007) 1. Unforeseen events upset me greatly2 (1–5)2.2 (0.9) 2. It frustrates me not having all the information I need3 (1–5)2.9 (1.0) 3. Uncertainty keeps me from living a full life2 (1–5)2.0 (1.0) 4. One should always look ahead so as to avoid surprises3 (1–5)3.1 (1.0) 5. A small, unforeseen event can spoil everything, even with the best of planning2 (1–5)1.8 (0.8) 6. When it’s time to act, uncertainty paralyses me2 (1–5)1.9 (0.9) 7. When I am uncertain I can’t function very well2 (1–5)2.2 (0.9) 8. I always want to know what the future has in store for me2.5 (1–5)2.6 (1.1) 9. I can’t stand being taken by surprise1 (1–5)1.5 (0.8) 10. The smallest doubt can stop me from acting1 (1–5)1.7 (0.9) 11. I should be able to organize everything in advance2 (1–5)2.3 (0.9) 12. I must get away from all uncertain situations2 (1–5)1.8 (0.9)II. Tolerance for Ambiguity (TFA) (Geller et al., 1993) 1. It really disturbs me when I am unable to follow another person’s train of thought4 (1–6)3.4 (1.2) 2. If I am uncertain about the responsibilities involved in a particular task, I get very anxious4 (1–6)3.5 (1.2) 3. Before any important task, I must know how long it will take2 (1–6)2.6 (1.1) 4. I don’t like to work on a problem unless there is a possibility of getting a clear-cut and unambiguous answer2 (1–6)2.6 (1.0) 5. The best part of working on a jigsaw puzzle is putting in that last piece4 (1–6)4.0 (1.4) 6. I am often uncomfortable with people unless I feel that I can understand their behavior2.5 (1–6)2.7 (1.2) 7. A good task is one in which what is to be done and how it is to be done are always clear4 (1–6)4.0 (1.3)III. Pearson Risk Attitude (PRA) (Pearson et al., 1995) 1. I enjoy taking risksa4 (1–6)4.0 (1.2) 2. I try to avoid situations that have uncertain outcomesb3 (1–6)3.3 (1.1) 3. Taking risks does not bother me if the gains involved are higha4 (2–6)3.8 (1.2) 4. I consider security an important element in every aspect of my lifeb4 (1–6)4.2 (1.1) 5. People have told me that I seem to enjoy taking chancesa4 (1–6)4.2 (1.3) 6. I rarely, if ever, take risks when there is another alternativeb3 (1–6)3.4 (1.2)IV. Ambiguity Aversion in Medicine (AAM) (Han et al., 2014) 1. I would not have confidence in a medical test or treatment if experts had conflicting opinions about it4 (1–6)3.5 (1.1) 2. Conflicting expert opinions about a medical test or treatment would make me upset2 (1–6)2.7 (1.1) 3. I would not be afraid of trying a medical test or treatment even if experts had conflicting opinions about ita,b4 (1–6)3.7 (1.1) 4. If experts had conflicting opinions about a medical test or treatment, I would still be willing to try ita,b3 (1–6)3.4 (1.1) 5. I would avoid making a decision about a medical test or treatment if experts had conflicting opinions about it.3 (1–6)3.4 (1.0)V. Physicians’ Reactions to Uncertainty (PRU) – Anxiety due to Uncertainty (Gerrity et al., 1995) 1. I usually feel anxious when I am not sure of a diagnosis4 (1–6)3.5 (1.2) 2. I find the uncertainty involved in patient care disconcerting4 (1–6)3.4 (1.3) 3. Uncertainty in patient care makes me uneasy4 (1–6)3.4 (1.3) 4. I am quite comfortable with the uncertainty in patient carea,b4 (1–6)3.6 (1.2) 5. The uncertainty of patient care often troubles me3 (1–6)3.1 (1.2)a These items were inverted prior to analysesb These items are scored inversely in the questionnaire Descriptive statistics of items in included instruments, n=202 a These items were inverted prior to analyses b These items are scored inversely in the questionnaire Exploratory factor analysis EFA identified three solutions with 5, 6 and 9 dimensions depending on extraction method. As excess dimensions were clearly indicated, the solution with 5 dimensions demonstrated stable structure regardless of rotation method (see Table 3 for solution with promax rotation due to correlations among factors). All items in PRU loaded on one factor (F1), which explained most of the variance (50.4%). AAM was stable, with all items loading on F5, but explained little total variance (6.4%). Cronbach’s alpha showed acceptable internal consistency for all five factors, but omega coefficient showed questionable internal consistency for F2, F4 and F5. Table 3Results of factor analyses; principal factor extraction method with Promax rotation Item a F1 F2 F3 F4 F5 IUS I.10.23 0.39 0.06−0.08−0.06 I.20.16 0.21 0.040.01−0.03 I.30.24 0.44 0.15−0.050.04 I.4−0.160.10 0.45 −0.10−0.03 I.50.02 0.52 0.210.030.02 I.60.01 0.75 −0.150.02−0.09 I.70.00 0.73 0.000.00−0.02 I.80.160.06 0.57 0.01−0.05 I.9−0.01 0.41 0.34−0.01−0.02 I.100.01 0.62 0.090.010.11 I.110.060.12 0.52 −0.06−0.09 I.120.25 0.36 0.220.030.13 TFA II.1 0.45 −0.020.250.05−0.30 II.2 0.63 0.180.06−0.02−0.18 II.3 0.38 0.050.160.13−0.06 II.40.090.18 0.27 0.070.24 II.5−0.06−0.06 0.51 0.100.26 II.6 0.44 0.050.110.11−0.08 II.70.210.02 0.40 −0.020.08 PRA III.10.06−0.020.02 0.84 −0.04 III.20.130.090.25−0.23 0.29 III.3−0.11−0.080.28 0.58 0.08 III.40.27−0.09 0.35 −0.250.13 III.50.060.04−0.13 0.85 0.13 III.6−0.06−0.070.20 0.51 0.15 AAM IV.10.07−0.080.180.05 0.56 IV.20.190.020.160.07 0.41 IV.3−0.02−0.040.170.08 0.64 IV.40.16−0.070.14−0.03 0.75 IV.50.01−0.120.230.02 0.33 PRU V.1 0.72 −0.08−0.01−0.080.07 V.2 0.81 −0.020.02−0.010.05 V.3 0.88 0.05−0.070.070.06 V.4 0.61 −0.090.190.13−0.18 V.5 0.70 0.06−0.12−0.010.13Variance explained (%)50.413.49.88.56.4Cronbach's alpha0.860.840.700.800.67Omega0.770.650.700.510.62IUS Intolerance of Uncertainty Scale, TFA Tolerance for Ambiguity, PRA Pearson Risk Attitude, AAM Ambiguity Aversion in Medicine, PRU Physicians’ Reactions to Uncertainty, F1-F5 refer to five factors and black bold formatting indicates loadings on specific factorsaItem numbering corresponds to the numbering in Table 1 Results of factor analyses; principal factor extraction method with Promax rotation IUS Intolerance of Uncertainty Scale, TFA Tolerance for Ambiguity, PRA Pearson Risk Attitude, AAM Ambiguity Aversion in Medicine, PRU Physicians’ Reactions to Uncertainty, F1-F5 refer to five factors and black bold formatting indicates loadings on specific factors aItem numbering corresponds to the numbering in Table 1 Cluster analysis In the steps below, cluster analysis with 2, 3 and 4 clusters was considered after inspection of the dendrogram. Step 1 – PRU alone (all other scales excluded) One cluster consisting of 58 cases with high UT (meaning low score) was stable, i.e. the same in all three solutions. The cluster with 36 cases with low UT was also the same in the solutions with 3 and 4 clusters. Cluster 1 and 2 for some items and 1 and 3 for other items in a 4-cluster solution were similar. Therefore, we concluded with the solution of 3 clusters, low UT (average silhouette 0.67, Kendall’s W 0.74), intermediate UT (average silhouette 0,20, Kendall’s W 0.10), and high UT (average silhouette 0.66, Kendall’s W 0.65). Step 2 – All scales One cluster consisting of 54 cases with low UT was stable. A cluster with 67 cases with intermediate UT was stable in solutions with 3 and 4 clusters. Since cluster 1 and 2 in a 4-cluster solution were very similar, we proceeded with the solution of 3 clusters. The average silhouette values were rather low (0.19, 0.28, 0.14 in low UT, intermediate UT and high UT clusters, respectively), even though Kendall’s W indicated a good consistency (respectively, 0.40, 0.51 and 0.41). Descriptive statistics within the clusters are presented in Table 4. Table 4Results of cluster analysis for PRU scale only and for all scales, n = 202. Numbers are means and standard deviationsItem1Only PRU (step 1)Low UT(Cluster 3, n = 36)Intermediate UT(Cluster 1, n = 108)High UT(Cluster 2, n = 58)p-value for ANOVA2 V.1 V.2 V.3 V.4 V.54.6 (0.7)4.8 (0.7)5.1 (0.7)5.2 (0.5)4.8 (0.6)3.6 (1.1)3.7 (0.9)3.7 (0.8)3.7 (0.9)3.2 (0.8)2.5 (0.8)2.0 (0.6)2.0 (0.7)2.5 (0.8)2.0 (0.6)* for all pairs* for all pairs* for all pairs* for all pairs* for all pairs Total mean4.9 (0.5)3.6 (0.5)2.2 (0.5) All scales (step 2) Low UT(Cluster 3, n = 54)Intermediate UT(Cluster 2, n = 67)High UT(Cluster 1, n = 81)p-value for ANOVA2 IUS I.1 I.2 I.3 I.4 I.5 I.6 I.7 I.8 I.9 I.10 I.11 I.123.0 (1.0)3.4 (1.1)2.9 (1.1)3.5 (0.9)2.5 (0.9)2.5 (1.2)2.8 (1.0)3.4 (1.1)2.0 (1.0)2.4 (1.1)2.9 (0.8)2.6 (0.9)2.0 (0.7)2.9 (0.9)1.9 (0.7)2.8 (0.9)1.5 (0.5)1.8 (0.7)2.2 (0.7)2.3 (0.8)1.5 (0.6)1.6 (0.6)2.1 (0.7)1.8 (0.7)1.8 (0.7)2.7 (1.0)1.4 (0.8)3.0 (1.0)1.4 (0.7)1.6 (0.7)1.8 (0.7)2.2 (0.9)1.2 (0.5)1.2 (0.4)1.9 (0.9)1.4 (0.6)*1–3, *2–3*1–3, **2–3*1–2, *1–3, *2–3*1–3, *2–3*1–3, *2–3*1–3, *2–3**1–2, *1–3, *2–3*1–3, *2–3*1–3, *2–3*1–2, *1–3, *2–3*1–3, *2–3*1–2, *1–3, *2–3Total mean2.8 (0.5)2.0 (0.3)1.8 (0.3) TFA II.1 II.2 II.3 II.4 II.5 II.6 II.74.2 (1.0)4.7 (0.8)3.2 (1.2)3.1 (1.1)4.5 (1.2)3.6 (1.2)4.8 (1.1)3.3 (1.1)3.4 (0.9)2.5 (1.0)2.9 (1.0)4.1 (1.4)2.5 (1.0)3.9 (1.1)3.0 (1.2)2.7 (1.0)2.2 (1.0)2.1 (0.8)3.6 (1.4)2.4 (1.2)3.4 (1.4)*1–3, *2–3*1–2, *1–3, *2–3*1–3, *2–3*1–2, *1–3**1–2, *1–3*1–3, *2–3*1–3, *2–3 Total mean4.0 (0.6)3.2 (0.5)2.8 (0.5) PRA III.1 III.2 III.3 III.4 III.5 III.64.4 (1.2)3.9 (0.9)4.1 (1.3)4.9 (0.9)4.7 (1.4)3.8 (1.4)3.8 (1.2)3.5 (0.9)3.6 (0.9)4.1 (1.0)4.0 (1.2)3.5 (1.1)3.9 (1.3)2.7 (1.0)3.8 (1.2)3.7 (1.1)4.2 (1.3)3.2 (1.0)**1–3, **2–3*1–2, *1–3**2–3*1–3, *2–3**2–3**1–3 Total mean4.3 (0.8)3.7 (0.7)3.6 (0.8) AAM IV.1 IV.2 IV.3 IV.4 IV.53.8 (1.2)3.3 (1.2)3.9 (1.1)3.4 (1.2)3.4 (1.1)3.7 (0.9)2.7 (1.0)3.8 (1.0)3.4 (0.9)3.7 (0.8)3.2 (1.0)2.3 (0.8)3.4 (1.0)3.3 (1.2)3.1 (1.0)**1–2, *1–3**1–2, *1–3, *2–3**1–3n.s.*1–2 Total mean3.6 (0.7)3.5 (0.6)3.1 (0.7) PRU V.1 V.2 V.3 V.4 V.54.4 (0.9)4.5 (0.9)4.6 (0.9)4.5 (0.9)4.0 (1.1)3.7 (0.9)3.8 (0.8)3.9 (0.8)3.9 (0.9)3.3 (0.9)2.6 (0.9)2.4 (1.0)2.3 (0.8)2.8 (1.1)2.3 (0.8)*1–2, *1–3, *2–3*1–2, *1–3, *2–3*1–2, *1–3, *2–3*1–2, *1–3, *2–3*1–2, *1–3, *2–3 Total mean4.4 (0.7)3.7 (0.6)2.5 (0.6)IUS Intolerance of Uncertainty Scale, TFA Tolerance for Ambiguity, PRA Pearson Risk Attitude, AAM Ambiguity Aversion in Medicine, PRU Physicians’ Reactions to Uncertainty1Item numbering corresponds to the numbering in Table 1; 2 The p-value based on ANOVA indicates the level of significance of the differences between pairs of clusters. The numbers represent the pairs of clusters (for example, 1–2 refers to Cluster 1 vs. Cluster 2), while the asterisks indicate the level of significance: *p < 0.01, ** p < 0.05 (n.s. refers to not significant) Results of cluster analysis for PRU scale only and for all scales, n = 202. Numbers are means and standard deviations V.1 V.2 V.3 V.4 V.5 4.6 (0.7) 4.8 (0.7) 5.1 (0.7) 5.2 (0.5) 4.8 (0.6) 3.6 (1.1) 3.7 (0.9) 3.7 (0.8) 3.7 (0.9) 3.2 (0.8) 2.5 (0.8) 2.0 (0.6) 2.0 (0.7) 2.5 (0.8) 2.0 (0.6) * for all pairs * for all pairs * for all pairs * for all pairs * for all pairs I.1 I.2 I.3 I.4 I.5 I.6 I.7 I.8 I.9 I.10 I.11 I.12 3.0 (1.0) 3.4 (1.1) 2.9 (1.1) 3.5 (0.9) 2.5 (0.9) 2.5 (1.2) 2.8 (1.0) 3.4 (1.1) 2.0 (1.0) 2.4 (1.1) 2.9 (0.8) 2.6 (0.9) 2.0 (0.7) 2.9 (0.9) 1.9 (0.7) 2.8 (0.9) 1.5 (0.5) 1.8 (0.7) 2.2 (0.7) 2.3 (0.8) 1.5 (0.6) 1.6 (0.6) 2.1 (0.7) 1.8 (0.7) 1.8 (0.7) 2.7 (1.0) 1.4 (0.8) 3.0 (1.0) 1.4 (0.7) 1.6 (0.7) 1.8 (0.7) 2.2 (0.9) 1.2 (0.5) 1.2 (0.4) 1.9 (0.9) 1.4 (0.6) *1–3, *2–3 *1–3, **2–3 *1–2, *1–3, *2–3 *1–3, *2–3 *1–3, *2–3 *1–3, *2–3 **1–2, *1–3, *2–3 *1–3, *2–3 *1–3, *2–3 *1–2, *1–3, *2–3 *1–3, *2–3 *1–2, *1–3, *2–3 II.1 II.2 II.3 II.4 II.5 II.6 II.7 4.2 (1.0) 4.7 (0.8) 3.2 (1.2) 3.1 (1.1) 4.5 (1.2) 3.6 (1.2) 4.8 (1.1) 3.3 (1.1) 3.4 (0.9) 2.5 (1.0) 2.9 (1.0) 4.1 (1.4) 2.5 (1.0) 3.9 (1.1) 3.0 (1.2) 2.7 (1.0) 2.2 (1.0) 2.1 (0.8) 3.6 (1.4) 2.4 (1.2) 3.4 (1.4) *1–3, *2–3 *1–2, *1–3, *2–3 *1–3, *2–3 *1–2, *1–3 **1–2, *1–3 *1–3, *2–3 *1–3, *2–3 III.1 III.2 III.3 III.4 III.5 III.6 4.4 (1.2) 3.9 (0.9) 4.1 (1.3) 4.9 (0.9) 4.7 (1.4) 3.8 (1.4) 3.8 (1.2) 3.5 (0.9) 3.6 (0.9) 4.1 (1.0) 4.0 (1.2) 3.5 (1.1) 3.9 (1.3) 2.7 (1.0) 3.8 (1.2) 3.7 (1.1) 4.2 (1.3) 3.2 (1.0) **1–3, **2–3 *1–2, *1–3 **2–3 *1–3, *2–3 **2–3 **1–3 IV.1 IV.2 IV.3 IV.4 IV.5 3.8 (1.2) 3.3 (1.2) 3.9 (1.1) 3.4 (1.2) 3.4 (1.1) 3.7 (0.9) 2.7 (1.0) 3.8 (1.0) 3.4 (0.9) 3.7 (0.8) 3.2 (1.0) 2.3 (0.8) 3.4 (1.0) 3.3 (1.2) 3.1 (1.0) **1–2, *1–3 **1–2, *1–3, *2–3 **1–3 n.s. *1–2 V.1 V.2 V.3 V.4 V.5 4.4 (0.9) 4.5 (0.9) 4.6 (0.9) 4.5 (0.9) 4.0 (1.1) 3.7 (0.9) 3.8 (0.8) 3.9 (0.8) 3.9 (0.9) 3.3 (0.9) 2.6 (0.9) 2.4 (1.0) 2.3 (0.8) 2.8 (1.1) 2.3 (0.8) *1–2, *1–3, *2–3 *1–2, *1–3, *2–3 *1–2, *1–3, *2–3 *1–2, *1–3, *2–3 *1–2, *1–3, *2–3 IUS Intolerance of Uncertainty Scale, TFA Tolerance for Ambiguity, PRA Pearson Risk Attitude, AAM Ambiguity Aversion in Medicine, PRU Physicians’ Reactions to Uncertainty 1Item numbering corresponds to the numbering in Table 1; 2 The p-value based on ANOVA indicates the level of significance of the differences between pairs of clusters. The numbers represent the pairs of clusters (for example, 1–2 refers to Cluster 1 vs. Cluster 2), while the asterisks indicate the level of significance: *p < 0.01, ** p < 0.05 (n.s. refers to not significant) Cluster analysis groups individuals according to questionnaire response patterns. The Sankey diagram (Fig. 1) illustrates that residents with high UT according to PRU with few exceptions also have high UT measured by all instruments together. Among those with low UT according to PRU, one third had high UT measured by all instruments, while among those with intermediate UT according to PRU roughly one half had low or high UT as measured by all instruments. Sankey diagram illustrating distribution of residents to clusters in step 1 (using PRU only) and step 2 (using all instruments combined) Table 5 shows that the clusters demonstrated a significantly higher prevalence of low UT among female residents. Further, there was also a significantly higher prevalence of low UT among residents having studied in non-Scandinavian speaking countries compared to residents having studied in Norway. Table 5Distribution of residents in clusters according to work experience, gender, and geographical region of university, using PRU alone (Step 1) or all instruments combined (Step 2). (n = 202)Step 1 Only PRULow UT(Cluster 3, n = 36)Intermediate UT(Cluster 1, n = 108)High UT(Cluster 2, n = 58)p-valueMonths of experience n3310555 mean (SD)8.5 (8.2)9.5 (7.6)10.5 (7.3)0.4871Gender, n (%) n3610857 female32 (88.9)75 (69.4)30 (52.6)0.0012Geographical region of medical university, n (%) n3610857 Scandinavia16 (44.4)66 (61.1)43 (75.4)0.0102 Outside Scandinavia20 (55.6)42 (38.9)14 (24.6) Step 2 All instruments Low UT(Cluster 3, n = 54)Intermediate UT(Cluster 2, n = 67)High UT(Cluster 1, n = 81)p-valueMonths of experience n536377 mean (SD)9.4 (8.3)8.3 (5.7)10.8 (8.4)0.1461Gender, n (%) n546780 female44 (81.5)47 (70.1)46 (57.5)0.0132Geographical region of medical university, n (%) n546780 Scandinavia27 (50.0)40 (59.7)58 (72.5)0.0272 Outside Scandinavia27 (50.0)27 (40.3)22 (27.5)1ANOVA2χ2-test Distribution of residents in clusters according to work experience, gender, and geographical region of university, using PRU alone (Step 1) or all instruments combined (Step 2). (n = 202) 1ANOVA 2χ2-test Discussion In this exploratory study we administered five brief UT measures to novice medical trainees, and found that the PRU—the only UT measure that is specific to physicians and medical care but not to any particular issue or source of uncertainty—demonstrated the most stable, consistent factor structure, accounted for the most variance in responses, and separated physicians most clearly into groups. However, the addition of the other UT measures—which are not specific to physicians and medical care, and focused on particular issues or sources of uncertainty—reclassifies many of those with low or intermediate UT according to PRU to higher levels of UT. These findings thus suggest that non-physician-specific instruments may have added value in discriminating between residents, and the observed factor loadings of individual items from these instruments suggest why this may be the case. For instance, F1 includes not only all items of the PRU but three items of the TFA that arguably all deal with individuals’ sense of unease or distress with the uncertainty involved in dealing with important tasks and/or people. F2, in contrast, includes only items from the IUS-12 that mainly comprise the “inhibitory” subscale tapping individuals’ sense of paralysis from uncertainty [27]. F3 includes items from the “prospective” subscale of the IUS-12, focused on a desire for predictability of the future, as well as items from the TFA and PRA that similarly focus on a desire for future control. Finally, F4 consists exclusively of PRA scale items focused on risk aversion, while F5 consists of the entire AAM scale and a single PRA scale item, all of which arguably focus on ambiguity aversion. Much more work is needed to confirm these factor loadings and their conceptual underpinnings; however, our study provides empirical evidence for why measuring these different facets of UT may be not only theoretically coherent but practically useful. For physicians or medical educators interested in assessing UT among residents, the PRU and non-physician-specific measures may identify subgroups of physicians who are predisposed to stress or other aversive psychological reactions to the uncertainties of clinical care, and who might benefit from targeted interventions. They may also help physicians identify specific types of uncertainty that they may be averse to, or specific types of uncertainty responses that they may be prone to, and to develop strategies for dealing with these responses. In a recent paper, we found that novice residents tended to use the terms uncertainty and insecurity interchangeably, and that lack of safety was a predominant factor in how they handled clinical situations [28]. Since so many of those with low or intermediate UT according to PRU have higher level of UT when other instruments are added, these non-physician-specific measures may also be useful not only in identifying residents who might benefit from targeted interventions, but in helping them better understand how different types of uncertainty affect their personal sense of security and safety [27]. The findings of the current study suggest that UT measures may help learners to reflect upon their own attitudes towards different types of uncertainty, develop greater self-insight into their responses to these uncertainties, and be proactive in monitoring and adjusting their future responses. Our findings might be a foundation for development of a streamlined instrument. More research is needed to determine which measures are most feasible, acceptable, and useful for these purposes; for now, the choice of measures depends on educators’ interests and motivation in exploring their value. UT measures could thus be a valuable coaching tool. They may ultimately help supervising and coaching physicians to address junior colleagues’ ability to deal with the stresses of medical uncertainty [29]. For example, educators could administer UT measures to trainees at the beginning of a clinical rotation as a way of getting acquainted and helping trainees to reflect upon their own approaches to uncertainty and to work with educators to set learning goals. We are currently conducting a pilot project with new Emergency Medicine residents that employs this current battery of UT measures in this way. We are finding that these measures serve as good conversation starters that also allow teachers to share their own experiences and wisdom about managing uncertainty, and to build productive relationships with learners, as well as providing a foundation for focused discussions of strategies for coping with uncertainty in clinical care. UT measures could also be readministered during or at the end of clinical rotations, to foster further reflection on learners’ growth in managing uncertainty. We are also pursuing this strategy in our own research. UT measures could also be useful to “screen” for learners at greater risk of negative responses to uncertainty and to provide targeted UT-fostering interventions for them. Such interventions remain to be developed, but might include small group discussions and dedicated debriefing activities focused specifically on creating a safe space for learners to share and reflect on their experiences with uncertainty, as well as individual counseling and feedback. More research is needed not only to develop such interventions but to train faculty to implement them and to provide support to cultivate learners’ uncertainty tolerance. As the PRU is physician-specific and asks questions concerning diagnostic and therapeutic decisions – decisions that medical students normally do not have the full responsibility for – one could argue that the non-physician-specific measures might be a better way to map UT in medical students, at least in their first years. We found that low UT was significantly more common among female than male residents. This finding echoes results in a recent large US study [30]. In Norway, females now comprise the majority of medical students. Male supervisors might not be sufficiently aware of the specific needs of many female residents. As low UT is associated with burnout [31], lack of focus on these needs might worst case lead to a lower system work capacity. First year residents in Norway do not choose where to start training, so findings cannot be attributed to where female and male residents decide to work. We also found lower UT among residents trained in non-Scandinavian speaking countries, even if most were fluent in Norwegian. With few exceptions, these residents’ university curricula were taught in English, while patients they met then rarely were fluent in English. Language is a critical source of understanding [32]. In an experimental study in the US even a slight change of English words in handoffs led to significantly higher anxiety due to uncertainty [33]. In a world with high migration, its effect on physicians’ professional identity should be recognized [34]. One reason for the finding could be that there is a selection difference between those trained abroad vs. in Norway, but we do not have data to support or explore this. Finally, there is an established relationship between age and UT, however these observations derive from studies that include a wide age range, either in large surveys [35, 36] or specific groups of physicians [37]. In our age-wise quite homogenous group of 1 st year residents, we considered months of clinical work experience to be a more relevant factor to study. We think our paper sheds light on subtle effects of contextual factors on uncertainty and warrants a critical eye on generalizability of findings in this and later studies of UT in clinical settings. This study has important limitations. We did not assess correlates of UT—including physicians’ actual behaviors or responses to uncertainty in clinical care settings, or the well-being of either physicians or their patients—raising questions about the significance of differences in UT and the value of assessing them. Conclusion Uncertainty tolerance (UT) measures administered to 1 st year residents at commencement of their training differ in their performance. A physician-specific measure focused on negative emotional responses to uncertainty appears to best discriminate between individuals, although non-physician-specific measures that assess specific dimensions of UT add discriminatory power for residents with lower UT. UT measures may be useful in medical education as coaching tools that could help learners reflect on their own responses to uncertainties in clinical care, and to better manage these responses and prevent adverse outcomes. Supplementary Information Supplementary Material 1 Acknowledgements No acknowledgements. Abbreviations All abbreviations in the manuscript are spelt out when first presented. Abbreviations Ambiguity Aversion in Medicine scale Analysis of Variance Exploratory Factor Analysis Intolerance of Uncertainty Scale Pearson Risk Attitude scale Physicians’ Reactions to Uncertainty scale Standard Deviation Tolerance For Ambiguity scale Uncertainty Tolerance Author contributions PG conceived the idea, PG. and EHO. collected the data, JSB did the statistical analysis, all authors contributed to writing and editing the manuscript and take public responsibility for the content. Funding Open access funding provided by University of Oslo (incl Oslo University Hospital). No external funding. Data availability All data are available from the corresponding author (P.G.) or the statistician (J.S.B.) on reasonable request. Declarations Ethics approval and consent to participate The study was exempt from review by the Regional Medical Ethics Committee of Southeast Norway, application #273339. The Data Protection Officer at Akershus University Hospital has approved the way we store data (21/05892-6). Competing interests The authors declare no competing interests. Footnotes The original online version of this article was revised: the authors would like to request corrections in the Tables which were erroneously captured by the production team. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Change history 3/14/2026 The original online version of this article was revised: the authors would like to request corrections in the Tables which were erroneously captured by the production team. Change history 3/14/2026 A Correction to this paper has been published: 10.1186/s12909-026-08976-4 References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All data are available from the corresponding author (P.G.) or the statistician (J.S.B.) on reasonable request.